Are ssd's more fragile than hdd's?

[Thomas]

Recently my ssd died on me, and I've only had it since November. Now I use my computer about 12-14 hours a day so it gets alot of use. With my older computer which only had an hdd I used it for the same amount of hours a day, but for like 4 years with no problems. Was my ssd dying just a freak event or are they less durable than hdd's?

[tacoshy]

defraging doesnt make block writable again. defragging moves blocks clsoe togetehr which is not needed for the way a SSD works...

and formatting an entire SSd is not needed when youc an simply delete the content unless you want to clean install windows.

[Omega]

引用自 vadim：

引用自 Omega：

It doesn't matter to it where the data is, it can instantly access it no matter where the data is located.

You're wrong. But this is common misconception. Occurring from a misunderstanding of how SSD works.
Look at Intel site: https://ark.intel.com/products/94438/Intel-SSD-E-5400s-Series-48GB-M_2-80mm-SATA-6Gbs-16nm-TLC
They specify the performance for access within 8 GB of address space. Why? Because drive keeps addresses for 8GB LBA in the single block.

Ok, and will it have any meaning full impact on performance?

If I need to wait any less then 10ms longer for my SSD to access the data in another block I can't be bothered.

[vadim]

引用自 tacoshy：

defragging moves blocks clsoe togetehr which is not needed for the way a SSD works

Why do you think so? Can you explain your point of view? Answer the simple question: why Intel indicates performance for a limited range of addresses if access to any address, in your opinion, occurs equally fast?

引用自 Omega：

If I need to wait 10ms longer for my SSD to access the data in another block I can't be bothered.

Of course delay will be much less than 10ms. But I'm talking about the general principle.

[tacoshy]

for a HDD you have blocks writen on certain parts of a disk. The disk needs to spin so that the reading arm can acess it. If the data is split on 2 block on different sectors the disk has to spin more and the arm to move more to acess both block and read the data out of it. That take time and defragging aka bringing it closer to shortens the need tos pin the disk or to move the arm causing less read time.


For a SSd each block can be read in less then 20ns ... not ms but ns. so if you move the blocks closer to each other to acess them now in 18ns instead of 20ns is not worth the loss of TBW.

[TehSpoopyKitteh]

引用自 Thomas：

Recently my ssd died on me, and I've only had it since November. Now I use my computer about 12-14 hours a day so it gets alot of use. With my older computer which only had an hdd I used it for the same amount of hours a day, but for like 4 years with no problems. Was my ssd dying just a freak event or are they less durable than hdd's?

In terms of forensics, yes. Hard disk platters are way harder to destroy due to their having a protective layer of translucent cobalt protecting them. Even rust (in the form of oxidized cobalt) cannot possibly remove data from it as a result.

SSD's on the other hand, if water gets into any of the NAND Flash chips, you can say goodbye to any usefulness out of it.

[tacoshy]

引用自 The Spoopy Kitteh：

引用自 Thomas：

Recently my ssd died on me, and I've only had it since November. Now I use my computer about 12-14 hours a day so it gets alot of use. With my older computer which only had an hdd I used it for the same amount of hours a day, but for like 4 years with no problems. Was my ssd dying just a freak event or are they less durable than hdd's?

In terms of forensics, yes. Hard disk platters are way harder to destroy due to their having a protective layer of translucent cobalt protecting them. Even rust (in the form of oxidized cobalt) cannot possibly remove data from it as a result.

SSD's on the other hand, if water gets into any of the NAND Flash chips, you can say goodbye to any usefulness out of it.

same for the hdd but for the reason that normal water has salt into it which is small enough to get through the tiney air hole in a HDD which will cause a nice scratch on the disk that can eb scratched by even dust and completly destroy a disk.

[vadim]

引用自 tacoshy：

For a SSd each block can be read in less then 20ns ... not ms but ns. so if you move the blocks closer to each other to acess them now in 18ns instead of 20ns is not worth the loss of TBW.

You need to make two more steps. First step is simple: compare SSD with RAM. When you access RAM you cannot read single variable from it to CPU register. You always need to read whole cache line. For Core i its 64 bytes long. So you can read second double word extremely fast - in 4 cycles. But random access could be about 60 times slower. You need to flush another cache line, fill it with burst reading from RAM and so on. The same is true about SSDs. Models which I know performs prefetching from NAND to drive RAM. Because of that reading from following LBAs could be performed significally faster.

Second step is slightly more complicated. To begin with, it should be noted that the SSD memory is associative. The block number for it is just a tag pointing to the NAND page. These tags themselves are also stored in the NAND and read from there as necessary.
Here again you can draw a parallel with RAM.
To read a random RAM address, you need to translate the linear address into a physical address. But for such a translation, you must first read several (up to 5) levels of the page table (TLB). Therefore, access to a random RAM address is much slower than with sequential access.

[Omega]

引用自 vadim：

Of course delay will be much less than 10ms. But I'm talking about the general principle.

If we are talking micro or nanoseconds here it's not worth the time (10 whole seconds of my life to start the defrag) nor the write cycles to defrag. And even then, defragging is going to guarantee the data to be in the same block.

[TehSpoopyKitteh]

引用自 tacoshy：

引用自 The Spoopy Kitteh：

In terms of forensics, yes. Hard disk platters are way harder to destroy due to their having a protective layer of translucent cobalt protecting them. Even rust (in the form of oxidized cobalt) cannot possibly remove data from it as a result.

SSD's on the other hand, if water gets into any of the NAND Flash chips, you can say goodbye to any usefulness out of it.

same for the hdd but for the reason that normal water has salt into it which is small enough to get through the tiney air hole in a HDD which will cause a nice scratch on the disk that can eb scratched by even dust and completly destroy a disk.

It's a lot harder to recover data from a damaged SSD.

[Bad 💀 Motha]

I've defragged my SSDs since SSDs were a thing, it doesn't harm them.
Overall that is not really something you even need to make folks aware of anymore.
Most users never defrag their PC Drives, its why over the years most HDD users had such slow downs from time to time. And modern OS' will ensure your SSDs are never defragged in the old traditional manner, instead they perform a different defrag (aka Trim Optimization)

[oobymach]

SSD's have limited write lifetimes, the more you write to it (eg using windows paging file) the quicker it will die. Some ssd's can last 3000 rewrites per block, others like samsungs entire lineup are only guaranteed for 75 block rewrites. Years of internet can degrade an ssd and kill parts of the drive but the rest is usually fine for years afterward. I have an ssd with such damage and use it for a game drive and all the games on it work fine.

In terms of fragility vs hdd they're a solid state part so not only can they take more physical trauma theoretically they can last a million years if you never write to it.

[Omega]

Yet still most SSDs will outlive any HDD by decades.

[TehSpoopyKitteh]

引用自 Omega：

Yet still most SSDs will outlive any HDD by decades.

Well yeah...everyday use they're harder to ruin unless you defragment one...

If you use a hard disk drive constantly, you only have to replace actuators, belts, and/or motors every few decades or so...not to mention they cost $0.009 at the most per GB...

Now, as far as shock absorption, SSD's have less a chance than old HDD's in becoming a brick. Modern HDD's have a magnetic field that separates the actuator arm's read head from the surface of the platter...and that entire actuator arm is held in place by a simple neodymium magnet.

[tiger305]

well, I have a crucial 512 ssd, that's 3 years old and never defraged ..

2 months ago, I decided to check the drive, the ssd was 60% fragmented,
so, ran win defrag...

I checked the health of the drive before, after and even just now, according to
crucial software drive is still in good health , no errors, and 0% life used.....

I think I'll be defraging my ssd's once a year or until I can really see
negative results of defraging ssd occasionally ,,,,

oh, and 7 TB have been written to this crucial drive so far......

[Bad 💀 Motha]

Well here's some data for you, have at it and let me know where it might stand in the mix of all of this... and this is nothing but the truth...

I've used a few Samsung 850 EVO over the years (to basically endurance test, but not being too harsh on them); now I'll just use 2 of these drives as an example of my many tests; 1x 250GB and 1x 500GB; both have had over 80TB worth of re-writes in under 3-6 months; I did old-fashioned Defrag to them daily (using both Auslogics Disk Defrag and Piriform Defraggler), yes daily... to add to that stress and see how quickly they "might" wear out due to this type of usage, perhaps by a user with heavy work loads for example who might be able to put such drives through such stress in a rather short amount of time. I also would perform "Free Space Wiping" using Piriform CCleaner... and on-top of that, never assigned any "Over-Provisioning" space, like Samsung suggested. The drives both had an OS and Games on them, one Win7 64bit, the other Win10 64bit. Both contained a 16GB (min & max) preset for a PageFile of their own.

all of this and they still to this day, perform like new with zero signs of % of life lost.